Re: [PATCH] xfs: revert to double-buffering readdir

[Timothy Shimmin <tes@sgi.com>]

Christoph Hellwig wrote:
> The current readdir implementation deadlocks on a btree buffers locks
> because nfsd calls back into ->lookup from the filldir callback.  The
> only short-term fix for this is to revert to the old inefficient
> double-buffering scheme.
> 

Probably why Steve did this: :)

xfs_file.c
----------------------------
revision 1.40
date: 2001/03/15 23:33:20;  author: lord;  state: Exp;  lines: +54 -17
modid: 2.4.x-xfs:slinx:90125a
Change linvfs_readdir to allocate a buffer, call xfs to fill it, and
then call the filldir function on each entry. This is instead of doing the
filldir deep in the bowels of xfs which causes locking problems.
----------------------------


Yes it looks like it is done equivalently to before (minus the uio stuff etc).
I don't know what the 7fff* masking is about but we did that previously.
I hadn't come across the name[] struct field before,
was used to name[0] (or name[1] in times gone by) but found that is
a kosher way of doing things too for the variable len string at the end.

Hmmm, don't see the point of "eof" local var now.
Previously bhv_vop_readdir() returned eof.
I presume if we don't move the offset (offset == startoffset) then
we're done and break out?
So we lost eof when going to the filldir in the getdents code etc...

--Tim

> This patch does exactly that and reverts xfs_file_readdir to what's
> basically the 2.6.23 version minus the uio and vnops junk.
> 
> I'll try to find something more optimal for 2.6.25 or at least find a
> way to use the proper version for local access.
> 
> 
> Signed-off-by: Christoph Hellwig <hch@lst.de>
> 
> Index: linux-2.6/fs/xfs/linux-2.6/xfs_file.c
> ===================================================================
> --- linux-2.6.orig/fs/xfs/linux-2.6/xfs_file.c	2007-11-25 11:41:20.000000000 +0100
> +++ linux-2.6/fs/xfs/linux-2.6/xfs_file.c	2007-11-25 17:14:27.000000000 +0100
> @@ -218,6 +218,15 @@
>  }
>  #endif /* CONFIG_XFS_DMAPI */
>  
> +/*
> + * Unfortunately we can't just use the clean and simple readdir implementation
> + * below, because nfs might call back into ->lookup from the filldir callback
> + * and that will deadlock the low-level btree code.
> + *
> + * Hopefully we'll find a better workaround that allows to use the optimal
> + * version at least for local readdirs for 2.6.25.
> + */
> +#if 0
>  STATIC int
>  xfs_file_readdir(
>  	struct file	*filp,
> @@ -249,6 +258,121 @@
>  		return -error;
>  	return 0;
>  }
> +#else
> +
> +struct hack_dirent {
> +	int		namlen;
> +	loff_t		offset;
> +	u64		ino;
> +	unsigned int	d_type;
> +	char		name[];
> +};
> +
> +struct hack_callback {
> +	char		*dirent;
> +	size_t		len;
> +	size_t		used;
> +};
> +
> +STATIC int
> +xfs_hack_filldir(
> +	void		*__buf,
> +	const char	*name,
> +	int		namlen,
> +	loff_t		offset,
> +	u64		ino,
> +	unsigned int	d_type)
> +{
> +	struct hack_callback *buf = __buf;
> +	struct hack_dirent *de = (struct hack_dirent *)(buf->dirent + buf->used);
> +
> +	if (buf->used + sizeof(struct hack_dirent) + namlen > buf->len)
> +		return -EINVAL;
> +
> +	de->namlen = namlen;
> +	de->offset = offset;
> +	de->ino = ino;
> +	de->d_type = d_type;
> +	memcpy(de->name, name, namlen);
> +	buf->used += sizeof(struct hack_dirent) + namlen;
> +	return 0;
> +}
> +
> +STATIC int
> +xfs_file_readdir(
> +	struct file	*filp,
> +	void		*dirent,
> +	filldir_t	filldir)
> +{
> +	struct inode	*inode = filp->f_path.dentry->d_inode;
> +	xfs_inode_t	*ip = XFS_I(inode);
> +	struct hack_callback buf;
> +	struct hack_dirent *de;
> +	int		error;
> +	loff_t		size;
> +	int		eof = 0;
> +	xfs_off_t       start_offset, curr_offset, offset;
> +
> +	/*
> +	 * Try fairly hard to get memory
> +	 */
> +	buf.len = PAGE_CACHE_SIZE;
> +	do {
> +		buf.dirent = kmalloc(buf.len, GFP_KERNEL);
> +		if (buf.dirent)
> +			break;
> +		buf.len >>= 1;
> +	} while (buf.len >= 1024);
> +
> +	if (!buf.dirent)
> +		return -ENOMEM;
> +
> +	curr_offset = filp->f_pos;
> +	if (curr_offset == 0x7fffffff)
> +		offset = 0xffffffff;
> +	else
> +		offset = filp->f_pos;
> +
> +	while (!eof) {
> +		int reclen;
> +		start_offset = offset;
> +
> +		buf.used = 0;
> +		error = -xfs_readdir(ip, &buf, buf.len, &offset,
> +				     xfs_hack_filldir);
> +		if (error || offset == start_offset) {
> +			size = 0;
> +			break;
> +		}
> +
> +		size = buf.used;
> +		de = (struct hack_dirent *)buf.dirent;
> +		while (size > 0) {
> +			if (filldir(dirent, de->name, de->namlen,
> +					curr_offset & 0x7fffffff,
> +					de->ino, de->d_type)) {
> +				goto done;
> +			}
> +
> +			reclen = sizeof(struct hack_dirent) + de->namlen;
> +			size -= reclen;
> +			curr_offset = de->offset /* & 0x7fffffff */;
> +			de = (struct hack_dirent *)((char *)de + reclen);
> +		}
> +	}
> +
> + done:
> + 	if (!error) {
> +		if (size == 0)
> +			filp->f_pos = offset & 0x7fffffff;
> +		else if (de)
> +			filp->f_pos = curr_offset;
> +	}
> +
> +	kfree(buf.dirent);
> +	return error;
> +}
> +#endif
>  
>  STATIC int
>  xfs_file_mmap(
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/